vRA services deployment fails while trying to start one or more infrastructure services

Products

• VMware Aria Suite

Issue/Introduction

Symptoms:

• While deploying vRA services, the /opt/scripts/deploy.sh fails with the following error:

Error: Could not get apiVersions from Kubernetes: unable to retrieve the complete list of server APIs: metrics.k8s.io/v1beta1: the server is currently unable to handle the request

• The failure might happen with any infrastructure service e.g. symphony-logging, identity-service and others.
• The problem typically occurs during upgrade. Once the problem is rectified, the upgrade must be resumed.

Environment

VMware vRealize Automation 8.x
VMware vRealize Automation 8.1.x

Cause

• Kubernetes metrics server becomes unresponsive and cannot self-recover.

Resolution

To resolve the issue, upgrade to the next minor release available.Currently, there is no resolution available in vRA 8.0.1 and 8.1.

Workaround:
To workaround the issue, you can stop the metrics server. To do so, execute the command from shell:

kubectl delete -f /etc/k8s-config/metrics-server/

In cluster environments, this must be executed on one of the nodes.

After applying this KB, kubectl top nodes, kubectl top pods, and vracli service status commands will stop showing information about the CPU and memory loads. This information will still be available by the top and ps commands.